Significance of Northern Andes Terrane Extrusion and Genesis of the Interandean Valley: Paleomagnetic Evidence From the “Ecuadorian Orocline”

GPS data suggest that the NW South America corner forms a semi‐rigid block drifting NE‐ward along the regional dextral strike‐slip faults that bound an oceanic terrane accreted in Late Cretaceous times to W Ecuador and Colombia. However, the relevance of both strike‐slip versus thrust tectonics during Cenozoic times and their relation with oceanic terrane accretion are unclear. Here we report on the paleomagnetism of 31 mid‐upper Eocene to upper Miocene mainly volcanic sites from the Cordilleras Occidental and Real of southern Ecuador. Eleven sites show that the western Cordillera Occidental underwent a 24° ± 10° clockwise (CW) rotation with respect to South America after late Miocene times, while no rotation occurred further east. We relate the regional CW rotation to the emplacement of the Cordillera Occidental nappe onto the continental sediments of the Interandean Valley, blanketing the Cordillera's eastern margin. As rotation and continental sedimentation onset ages are similar, we interpret such tectonic depression as a narrow flexural basin formed ahead of the advancing nappe front. The 20°–30° CW Neogene rotation of the Cordillera Occidental is indistinguishable from the post‐Cretaceous rotation of the Coastal forearc oceanic blocks, implying that the whole W Ecuador Andean chain was detached and rotated over a mid‐crustal detachment during the last 10 Ma. Eocene‐Miocene paleomagnetic inclination values are systematically consistent with those expected for South America, thus excluding latitudinal terrane drift. We suggest that the Andes of Ecuador and Peru form the “Ecuadorian Orocline”, formed by opposing orogenic rotations around the Amazonian craton indenter.

. The chemical composition and age similarity of oceanic crust from Ecuador (and Colombia) coastal terranes and cuttings from ODP wells drilled within the Caribbean plate led to the widely accepted suggestion that the W Ecuador-Colombia terranes are fragments of the Caribbean plate left behind during its drift from an original paleo-Pacific location (Kerr et al., 1997(Kerr et al., , 2002Kerr & Tarney, 2005;Luzieux et al., 2006;Massonne & Toulkeridis, 2012;Spikings et al., 2001;Vallejo et al., 2009). Preliminary paleomagnetic data from the W Colombia Cordillera (Hincapié-Gómez et al., 2018;Montes et al., 2005), and the robust data set from Coastal Ecuador terranes (Luzieux et al., 2006;Roperch et al., 1987) were used to suggest a Caribbean origin at shallow southern Pacific latitudes and subsequent northward drift.
In this framework, the following issues are still unclear: (a) did the Northern Andean Block kept drifting northward during the Cenozoic? (b) Was such drift accommodated by individual faults or did it involve the northern Andes as a whole (e.g., Jiménez et al., 2014)? (c) How significant was strike-slip tectonics when compared to thrust tectonics during Cenozoic times?
Here we report on the first paleomagnetic data set from the Cordilleras Occidental and Real of Ecuador and attempt to untangle the relative contribution of thrust versus strike-slip tectonics in the development of the Ecuadorian Andes during mid-late Cenozoic times.
Our results indicate that no significant Cenozoic strike-slip deformation occurred in post-Eocene times and that the Cajas Plateau underwent a post-late Miocene 20°-30° CW rotation related to rotational thrust sheet emplacement west of the Azogues fault.

Geodynamic Context
The Ecuadorian Andes belong to the Northern Andean chain (Figure 1). Compared to the high plateaus of the Central Andes, they show lower mean elevation (∼2,500), narrower width, and occurrence of exotic terranes interpreted as oceanic crust blocks accreted to South America during late Cretaceous times (Aspden & Litherland, 1992;Luzieux et al., 2006;Ramos, 2009;Spikings et al., 2005Spikings et al., , 2010. The Northern Andes formed in a complex tectonic setting, characterized by the interaction of three main plates: South America, Nazca, and Caribbean ( Figure 1a). GPS data show that-considering a South America fixed reference frame-the Nazca plate moves eastward by 6 cm/yr and subducts below the Andes, whereas the Caribbean plate drifts toward the ESE at ca. 2 cm/yr ( Figure 1b).
Complex plate interaction gives rise to a wide deformation zone, extending from Ecuador to Colombia and Venezuela, characterized by both submarine/onshore accretionary wedges and seismically active dextral faults (Alvarado et al., 2016;DeMets et al., 2010;Mora et al., 2017;Pousse-Beltran et al., 2017;Trenkamp et al., 2002). After the study done by Pennington (1981), this Andean sector was considered as a proper microplate: the Northern Andean Block (NAB in Figure 1a), undergoing 0.6 cm/yr NNE-ward (Trenkamp et al., 2002) independent drift with respect to nearby plates (Cediel et al., 2003;Egbue & Kellogg, 2010;Ramos, 2009). The boundary between South America and NAB is assumed to coincide with both regional dextral strike-slip and reverse fault systems that extend from the Gulf of Guayaquil (Ecuador) to the gulf Triste (Venezuela), following main topographic ranges and changing strike and deformation style along their length (Figure 1; Baize et al., 2020;Egbue & Kellogg, 2010;Jiménez et al., 2014;Nocquet et al., 2014;Pousse-Beltran et al., 2017;Trenkamp et al., 2002). Active seismicity along both types of faults has been recorded and inferred for historical time (e.g., Beauval et al., 2010;Dimate et al., 2003;Paris, 2000). In addition, a compilation of dextral fault displacements from field constraints suggests variable slip rates from 0.4 to 1 cm/yr during the last 1.8 Ma (Baize et al., 2020;Egbue & Kellogg, 2010;Eguez et al., 2003;Pousse-Beltran et al., 2017). Dextral faults are assumed to have formed as Late Cretaceous suture zones, as oceanic terranes occur along the western NAB, and subsequently reactivated during Cenozoic times.

Regional Geological Setting and Features of the Study Area
From a tectonostratigraphic and geomorphologic point of view, the Ecuadorian Andes can be subdivided into five trench-parallel domains bounded by major faults (Figure 1). From West to East, the Coastal zone represents the fore-arc domain and is composed of upper Cretaceous mafic plume-related basement overlain by upper Cretaceous, and Paleocene to Miocene volcanoclastic sequences (e.g., Benitez, 1995;Gansser, 1973;Goossens & Rose, 1973;Jaillard et al., 1995Jaillard et al., , 1999. The Pallatanga fault marks the boundary between the Coastal fore-arc zone and the Cordillera Occidental, which consists of upper Cretaceous island arc volcanic rocks, oceanic-related fragments, plutons, and unconformable upper Cretaceous to Eocene turbidites. The Cordillera Occidental is intruded by middle Eocene to upper Cenozoic granites and is overlain by post-late Eocene continental arc volcanic rocks (  The Pallatanga fault represents the southern segment of the regional Dolores -Guayaquil dextral megashear, which extends from the Gulf of Guayaquil to the Romeral fault in Colombia (Figure 1b). In detail, at its southernmost end, it displays a WSW-ENE trend and is characterized by an extensional horsetail structure that induced the opening of a pull-apart basin since 1.8 Ma (Witt & Bourgois, 2010;Witt et al., 2006). Northward, the Pallatanga fault system changes strike to SW-NE and SSW-NNE, cutting the western edge of the Cordillera Occidental with several splay faults from 3° to 2° 30′S. Further north, the N-S Chimbo-Toachi shear zone connects the Pallatanga and the Romeral fault in Colombia (Figure 1b). The main Pallatanga strand extends along the Cordillera Occidental and at ∼1°N joins to the East the Cayambe-Afiladores fault system in southern Colombia (Figure 1b). Relying on morphologic evidence, mean Holocene dextral slip rates of 3.5-4 mm/yr were reported along a Pallatanga fault splay from the central Ecuadorian Andes (Winter & Lavenu, 1989;Winter et al., 1993).
The contact between the Cordillera Occidental and the Cordillera Real is marked by the Peltectec fault. The roots of a Triassic-Jurassic volcanic arc emplaced into metamorphosed Paleozoic marine and volcanic sequence are exposed along the axis of the chain, whereas the western Cordillera Real comprises Triassic and Jurassic submarine basaltic-andesitic volcanic series (Aspden & Litherland, 1992;Litherland et al., 1994;Pratt et al., 2005). The Peltectec fault runs parallel to the Pallatanga fault with a NE trend south of 2°S. It gently bends to the NW at 2°S and continues to the NE from 1°S northward. At ∼1°N it connects with the Cayambe-Afiladores fault system and with the eastern Andes front in Colombia ( Figure 1b). The fault has been considered as a late Jurassic (Aspden & Litherland, 1992;Litherland et al., 1994) or late Cretaceous (Spikings et al., 2001 shear zone that formed during the main deformation event of the Cordillera Real, after the accretion of the Cordillera Occidental block (Pratt et al., 2005). However, episodic exhumation events were inferred to cluster during the late Eocene, early Miocene, and Late Miocene to present (Spikings et al., 2001). Relying on thermochronologic data, the activity of the northern portion of the Peltectec fault has been inferred to span from 15 Ma to recent times (Spikings et al., 2010).
South of 2°S the Interandean Valley lies some 50 km E of the Pallatanga fault and covers the eastern portion of the Cordillera Occidental. Here the deformation is dominated by closely spaced NE-trending thrust faults ( Figure 2; Hungerbühler et al., 2002;Steinmann, 1997;Steinmann et al., 1999), and the Peltectec fault system and the Cordillera Real lie further East ( Figure 2).
Our study focuses on the Cordillera Occidental, the Interandean Valley, and the Cordillera Real of Ecuador, between 2°40′ and 3°S ( Figure 2). Here remnants of late Cretaceous oceanic terranes (Yunguilla and Pallatanga Formations in Figure 2) are overlain by Paleocene turbidites (Angamarca Group; e.g., Jaillard et al., 2004). Vigorous volcanic arc activity from the late Eocene (part of Quingeo and Chinchin units) to the early Miocene (Saraguro Fm.) yielded massive ignimbrite successions filling remnants of late Eocene and early Oligocene continental basins (part of Quingeo and Chinchin units (Hungerbühler et al., 2002;Steinmann, 1997; Figure 2). Particularly, the Saraguro Fm. largely outcrops in the area between the Pallatanga fault and the Cuenca city hereafter referred to as the Cajas Plateau. There, the welded early Miocene ignimbrite succession is generally sub-horizontal to gently tilted (Chiaradia et al., 2004;Mulas et al., 2017, Figure 3; Table 1).
Extensional tectonics occurring during the middle Miocene led to the formation of a coastal basin spanning the whole Ecuadorian Andes west of Cuenca (Steinmann, 1997;Hungerbühler et al., 2002). The sedimentary succession-preserved only in the Interandean Valley-consists of 2,700 m of fluvial, deltaic, brackish delta plain, lacustrine, and coastal sediments (Hungerbühler et al., 2002) firmly indicating that in mid-Miocene times the Cordillera Occidental was characterized by a coastal environment with marine influence (Bristow & Parodiz, 1982;Feldmann et al., 1993;Hungerbühler et al., 2002;Martínez-García et al., 2017;Nuttall, 1990;Steinmann, 1997). Steinmann et al. (1999) and Hungerbühler et al. (2002) proposed that during the mid-late Miocene this wide basin was inverted by E-W directed shortening, which also yielded the main exhumation event of the Cordillera Occidental.
During and after tectonic inversion, continental deposition took place only in the Interandean Valley that became the depocenter of sediments eroded from the Cordillera Occidental. A regional unconformity preserved in the Interandean Valley marks the transition from coastal deposits to continental sediments (Hungerbühler et al., 2002). The compressive deformation stage is well preserved in the Interandean depression by tilted and deformed Miocene coastal to continental strata ( Figure 4). Thrust faulting and folding are observed along the valley (Figure 4b and 4c). West of the Cojitambo lava dome the increase of strata dip toward the East additionally suggests progressive tilting approaching a thrust-fault (Figures 3c and 4a). Steinmann et al. (1999) suggested that the Cojitambo lava dome (8 Ma) emplacement postdated the main deformation event and that late Miocene-Pliocene alluvial and pyroclastic air fall deposits seal the deformed basin fill.
To the East of the Interandean Valley, the Peltectec fault cuts Jurassic basalts and its trace fades out below Eocene to early Miocene volcanic ( Figure 2). Finally, the Subandean zone to the East of the Cordillera Real is composed of steep east-verging thrust fronts forming the presently active Andean front (Legrand et al., 2005). Here compressive tectonics started in the Miocene and continued to present times (e.g., Baby et al., 2013;Roddaz et al., 2010).

Paleomagnetic Sampling and Methods
We aimed at paleomagnetically characterizing the Cordillera Occidental, the Interandean Valley, and the Cordillera Real, and constraining the rotations related to the main faults separating them. Consequently, samples were collected along an NW-SE transect, roughly from the Pallatanga to the Peltectec faults SIRAVO ET AL.     Table 1 Paleomagnetic Directions From the Cordilleras Occidental and Real of Ecuador    (Table 1). Hungerbühler et al. (1995), Steinmann (1997), Steinmann et al. (1999), and Hungerbühler et al. (2002) dated the igneous lithologies and tephra layers interbedded in the sedimentary succession along the sampling area using apatite (AFT) and zircon fission tracks (ZFT) analysis, as listed in A total number of 300 paleomagnetic samples were gathered at 31 sites ( Figure 2 and Table 1). At each site, we drilled eight-ten cores using a petrol-powered portable drill cooled by water-except site ECU13 that was hand sampled in soft upper Miocene lacustrine sediments-and oriented them in situ using both a Sun and a magnetic compass, corrected for the local geomagnetic declination in September/October 2019 (∼3°W according to NOAA's National Geophysical Data Center https://www.ngdc.noaa.gov/geomag/calculators/magcalc.shtml). The cores were then cut into standard paleomagnetic specimens of 22 mm height.
Paleomagnetic measurements were performed in a shielded room, using a 2G Enterprises direct current-superconducting quantum interference device cryogenic magnetometer. Alternating field (AF) demagnetization using three coils online with the magnetometer starting from 10 mT and reaching a maximum AF peak of 120 mT (10 mT steps until 80 and 20 mT steps afterward) was routinely used in 10 steps. Fifty-eight samples from six sites that were not demagnetized at 120 mT (Table 1), and samples from site ECU21, were thermally cleaned using a Pyrox shielded oven in 10 steps up to 690°C.
Demagnetization data were plotted on orthogonal vector component diagrams (Zijderveld, 1967), and principal component analysis was used to isolate magnetization components (Kirschvink, 1980). Site-mean paleomagnetic directions were computed using Fisher (1953) statistics and plotted on equal-area projections, using the Remasoft software. Paleomagnetic rotation and flattening values with respect to the South America plate were calculated according to Demarest (1983), using reference paleo-poles from Torsvik et al. (2012). Rotations were calculated for both site-mean (Table 1) and locality-mean directions. Mean ages of the sampled rocks as derived from absolute and/or relative dating available from the literature were used to select the corresponding age paleopoles (Table 1).
The magnetic mineralogy was investigated for one sample per site by thermo-magnetic curves performed by an AGICO MK1-FA Kappabridge. The magnetic susceptibility of a powdered sample per site was measured in a heating-cooling cycle from room temperature up to 700°C.
We mostly sampled volcanics that are expected to record the paleo secular variation (PSV) of the geomagnetic field, and therefore to yield paleomagnetic directions more scattered than those from sediments. This represents a potential error source SIRAVO ET AL. and flattening (F) values, and relative errors ΔR and ΔF (according to Demarest (1983)) are relative to coeval D and I. for data interpretations in terms of tectonic rotations (e.g., Merrill et al., 1996). Therefore, following the method of Deenen et al. (2011) that considers the virtual geomagnetic pole (VGP) distribution and the relative statistical precision parameter K and A95 as a function of N (number of observations), we tested the PSV averaging of both site-mean and locality-mean paleomagnetic datasets (Tables S1 and S2). A95 values were obtained from Fisher statistics on VGPs calculated from magnetization components from specimens gathered in each site (site-mean A95), and given belt sectors (locality-mean A95).
Mean inclinations from data groups of the same age were used to calculate paleolatitudinal values.

Paleomagnetism and Evaluation of Latitudinal Drift
Scattered or erratic demagnetization diagrams were observed at seven sites that were excluded from further considerations (Table 1). In the remaining 24 sites, a viscous component was removed at 20 mT or 300 °C ( Figure 5). In the AF-cleaned sites, a characteristic magnetization component (ChRM) was isolated in the 20-120 mT field interval (Figure 5a-5d). In the thermally cleaned sites, a high-temperature (HT) component was identified in the 400-690°C temperature range (Figure 5e-5g). ChRMs and HT components were used to calculate site-mean paleomagnetic directions. Thermomagnetic curves ( Figure S1 in supporting information) show predominantly 550-600°C Curie temperatures, in some cases associated with minor susceptibility drop at 250°C. Demagnetization diagrams together with thermo-magnetic curves suggest magnetite and subordinate titano-magnetite (or pyrrhotite at sites ECU08, 17, and 22) as main magnetic carriers, along with hematite at the seven thermally demagnetized sites (ECU05, 10, 15, 21, 26, 28 and 31).
The α 95 values relative to the site-mean paleomagnetic directions vary from 3.2° to 20.9°, 10° on average, whereas sites ECU21 and ECU31 that exceed a 25° α 95 threshold value were discarded. The reliability of few sites showing α 95 values ranging from 18° to 21° is supported by a comparison of their A95 values with validity limits by Deenen et al. (2011) that will be discussed below. The reliable sites show both normal (12 sites) and reverse (10 sites) polarity states ( Figure 6 and Table 1).
The inclination-only (Arason & Levi, 2010)  nate (Tables S3-S5 in supporting information). However, the magnetic mineralogy is mostly dominated by magnetite, the occurrence of directions acquired on both polarity states, the spread of both in situ and tilt corrected paleomagnetic directions, and their distance from the local (normal-and reverse-polarity) geocentric axial dipole field directions ( Figure 6) imply no magnetic overprint and support the primary nature of the isolated ChRMs and HT components.
Reverse-polarity directions with southerly declinations were inverted to normal polarity before performing a directional analysis of site groups from given sectors of the chain.
Considering both in-situ and tilt-corrected sites to the west of the Azogues fault (ECU02-10; 12; 14-15 in Figure 2), the resulting mean direction is defined as D = 18.4°, I = −12.3° (α 95 = 12.6°). However, to avoid mixing in-situ and tilt-corrected sites we rely on the in-situ population only, yielding a mean direction given by D = 21.5°, I = 15.9° (α 95 = 11.6; Figure 6). This is used to evaluate the regional rotation pattern of the Cordillera Occidental (group "WEST" in supplementary tables and Figures 6 and 7b).
To the E of the Azogues fault, most of the directions are in in-situ coordinates, except for sites ECU20 and ECU19. However, tilt-corrected direction from site ECU20 and in-situ directions from nearby (2.5 km) site ECU17 and ECU16 are similar ( Figure 6) (Figure 7a and Table S1). The A95 from only six sites (ECU02,03,07,15,20,and,22) are located below the lower boundary, suggesting no PSV averaging. This result is somehow surprising, as volcanic sites drilled in single volcanic units should record a snapshot of geomagnetic field direction thus are expected to systematically not average the PSV. We conclude that in the majority of our volcanic sites we sampled, in fact, multiple volcanic units (i.e., multiple pyroclastics flows from the same formation) where PSV is averaged out. Sites were SIRAVO ET AL.

10.1029/2020TC006684
13 of 25 When sites are grouped in structurally homogeneous domains (i.e., no main fault occurs between them), the locality-mean directions show a complete PSV averaging (Figure 7b and Table S2), showing that our data can be used to assess tectonic rotations. Moreover, several works in the past proved that orogenic rotations can be evaluated by paleomagnetic data from volcanic rocks, provided that a sufficiently high number of data are averaged out (e.g., Andreani et al., 2014;Gattacceca et al., 2007;Hernandez-Moreno et al., 2014Pellegrino et al., 2018;Piper et al., 1997;Márton et al., 2006;Siravo et al., 2020, among many others).

Rotation Pattern and Tectonic Implications
Site-mean rotation values calculated using in-situ and tilt-corrected coordinates with respect to South America are both counterclockwise (CCW) and CW in sense, varying from −32° to 68° (Figure 2; Table 1; Figure S2 in supporting information). Locality-mean in-situ directions are used to evaluate regional rotation patterns. A rather uniform 23.9° ± 9.6° CW rotation (R) pattern characterizes the domain W of the Azogues fault, which includes the western Interandean Valley, the Cajas Plateau of the Cordillera Occidental, and the Pallatanga fault zone (Figures 2 and Figure S2). Here, the uniquely CCW-rotated site ECU04 is considered an outlier and discarded. To the E of the Azogues fault, sites ECU16, 17, and 20 are on average not rotated (R = −10.8° ± 20.5°). Sites ECU18, 19, 22, and 30 from along the Peltectec Fault provide a non-significant rotation R = 9.9° ± 19.1°. Sites ECU25, 26, straddling the Santa Ana dextral fault, are rotated CW by 68° ± 9.0° and 40° ± 15.0°, respectively. Finally, site ECU28 is far from any other site to infer a reliable rotation pattern and will not be considered onward (Figures 2 and S2).

Paleomagnetic Assessment of Post-Mid Eocene Strike-Slip Fault Offset
There is wide evidence that-in the vicinity of major strike-slip faults-the paleomagnetic pattern is characterized by rotated crustal blocks and secondary strike-slip faults separating them (e.g., McKenzie & Jackson, 1986;Ron et al., 1984;Sonder et al., 1994, among many others). It has been shown that-if a quasi-continuous crust deformation model is assumed-the width of the rotation-deformation zones adjacent to the faults and the maximum rotation values are proportional to strike-slip offset, thus can be used to estimate the significance of the faults themselves (Lamb, 1987;Nelson & Jones, 1987;Sonder et al., 1994). Major continental strike-slip faults with offsets of some tens of km (or more) were shown to generate a 10-20 km wide damage rotation zone on each fault side, with CW (CCW) rotations exceeding 90° in case of dextral (sinistral) faults (e.g., Hernandez-Moreno et al., 2014Pellegrino et al., 2018;K. Randall et al., 2011;Siravo et al., 2020;Sonder et al., 1994;Speranza et al., 2018).
Concerning our data set, four sites straddling the Peltectec fault (ECU18, 19, 22, and 30) are scattered and on average show no rotation (R = 9.9° ± 19.1°). Similarly, site ECU08, located in between the Pallatanga fault system shows a small 10° rotation, while the 28° ± 16° CW rotation of site ECU09 at 1 km from the fault is not significantly different from the regional 24° ± 10° CW rotation of the Cordillera Occidental. Nevertheless, a Quaternary-Holocene dextral strike-slip offset along the Pallatanga fault is reported (Winter & Lavenu, 1989;Winter et al., 1993). However, the Holocene displacements values estimated by Winter et al. (1993) (from 27 ± 11 m to 960 ± 70 m) translate to a barely detectable maximum paleomagnetic rotation ranging from 3° to 6°, when the formula of Lamb (1987) and widths of the deformation-rotation zone of 5, 10, and 20 km are assumed (e.g., Hernandez-Moreno et al., 2014;Pellegrino et al., 2018). Our data provide no evidence for a paleomagnetically-detectable offset along the Peltectec and Pallatanga fault systems after Middle Eocene and Early Miocene respectively, confirming previous suggestions that the accretion of oceanic terranes to the western Ecuador margin was completed by Late Cretaceous times (Luzieux et al., 2006;Spikings et al., 2010;Vallejo et al., 2009Vallejo et al., , 2019. Conversely, a dextral fault displacement may be inferred considering the systematic CW rotations observed at sites ECU25 and 26, which are located up to 3 km from the Santa Ana fault, for which both reverse and dextral displacements were suggested (Hungerbühler et al., 2002). Again, by using the formula by Lamb (1987) and considering 68° as the maximum rotation angle and 6 km as the width of the rotation domain (including both sides of the fault), we infer a 14 km offset for the Santa Ana fault. Although only two sites are obviously not enough to solidly constrain fault displacement, it is clear that this is not a major transcurrent feature of the Andes, as sites ECU16, 17, and 20, located 15 km further south and a few km W of the fault are not rotated (R = −10.8° ± 20.55°; Figure 2).
We conclude that strike-slip induced-rotations are almost absent in the study area. Such evidence is indeed consistent with the paleo-latitude data of Figure 8 that shows no significant difference from South America's expected values. Thus our data exclude the occurrence in the Ecuadorian Andes of terranes undergoing latitudinal drift and therefore indicate no significant orogen-parallel strike-slip activity, at least since midlate Eocene (∼40 Ma).
The sites from the Cajas Plateau are far from any surface fault, besides the Pallatanga fault. However, our data suggest a lack of post early Miocene strike-slip deformation. Compressive deformation is conversely documented in the upper Miocene sediments exposed in the vicinity of the Cojitambo and Azogues faults ( Figure 4).
The age of nappe emplacement-related CW rotation and the onset of Interandean Valley continental sedimentation also coincide suggesting a genetic link between the two processes. Continental deposition in the Cuenca area started in the Middle-Late Miocene (first continental deposits were dated at 12.1 ± 1.2 Ma by Steinmann et al., 1999), while CW rotation is constrained to the Late Miocene (younger rotated data are from the Cojitambo lava dome, dated at 7.8 ± 0.8 Ma by Steinmann et al., 1999). Therefore, we infer that after the Late Miocene times the Cajas Plateau and the Cordillera Occidental were carried on top SIRAVO ET AL.

10.1029/2020TC006684
16 of 25  Figure 1c) and paleomagnetic rotations with respect to South America considering our data and those by Roperch et al. (1987) and Luzieux et al. (2006). See text for explanation.
of a mid-crustal rotational detachment whose front coincides with the Azogues fault in the Interandean Valley ( Figure 9). We also propose that Interandean Valley developed as a relatively narrow flexural basin in front of the advancing Cordillera Occidental deformation front and was subsequently involved in the compression by foreland propagating thrust faults (Figure 9), consistently with previous interpretations by Hungerbühler et al. (2002), Lavenu et al. (1995), and Steinmann et al. (1999), and that deformation likely continued through the Pliocene. Elongated sedimentary basins lying on deformed orogens in front of thrust sheets and acting as sediment depocenters for a few million years are observed further south in the Andes (Sierra Pampeana; e.g., Strecker et al., 1989) and in Tibet (e.g., Mula basin; Todrani et al., 2020). On the other hand, classical foreland basins of orogenic fronts are hundreds of km wide and may contain huge packages of turbidites thousands of meters thick (e.g., Davis et al., 1996).
East of the Interandean Valley, the lack of significant CW rotation excludes any post-Eocene strike-slip displacement along the Peltectec fault that marks the suture zone between the Cordillera Occidental and Cordillera Real blocks (Figure 9). This conclusion is partly inconsistent with previous studies that, relying on thermochronologic data from the Cordilleras Occidental and Real, suggest Eocene tectonic oblique-slip reactivations within specific fault-bounded massifs (Spikings et al., 2001(Spikings et al., , 2010. The alternative hypothesis is that strike-slip shear along the Peltectec fault pre-dated the Mid-Late Eocene sites paleomagnetically sampled by us. Our data yield the first paleomagnetic constraints to the rotational pattern of the Cordillera Occidental and Real of Ecuador; however previous upper Cretaceous paleomagnetic data from the Coastal forearc investigated the deformation history of the oceanic blocks that represent the crystalline basement of Ecuadorian Andes from the forearc to the Interandean Valley (Luzieux et al., 2006;Roperch et al., 1987). We recalculated the paleomagnetic rotations with respect to the South American margin comparing the combined data set by Roperch et al. (1987) and Luzieux et al. (2006) with updated plate reference poles by Torsvik et al. (2012) (Figure 1c; Table S6 in supporting information). We find that the oceanic basement formations (Piñon and San Lorenzo; late Turonian -Coniacian) and the younger detrital Cayo formation (Santonian-Maastrichtian) yield CW rotation values of 66° ± 17° and 28° ± 9°, respectively. A 40°-50° CW rotation and complete incorporation of the Caribbean-related oceanic blocks to the South American margin at 70 Ma was suggested by Luzieux et al. (2006) (Figure 10a), but no explanation was given for the remaining 28° ± 9° post-Cretaceous rotations.
Here we stress that the 28° ± 9° post-Cretaceous CW rotation of the Coastal forearc is statistically indistinguishable from the 24° ± 10° Neogene CW rotation documented by us in the Cordillera Occidental and Interandean Valley. Thus we argue for a Neogene age of the CW rotation documented by Luzieux et al. (2006) in uppermost Cretaceous strata from the Coastal forearc (Figure 10b).
We suggest that a mid-crustal rotational detachment lying below both the Coastal forearc and the Cordillera Occidental, emerging along the Interandean Valley, would explain the paleomagnetic data set from the oce-SIRAVO ET AL. anic terrains of Ecuador (Figures 9 and 10b). The Late Miocene-Pliocene emplacement of such a regional rotational tectonic nappe is in agreement with both an Early-Late Miocene shortening event reported from the Cordillera Occidental, the Interandean valley, and the Oriente foreland basin (Baby et al., 2013;Roddaz et al., 2010;Spikings et al., 2010;Steinmann et al., 1999).
Shortening also peaks at maximum values at the Bolivian Orocline apex with respect to adjacent orogenic limbs (e.g., McQuarrie, 2002).
Further north, from 10°S to 5°N the Nazca trench bends westward forming a wide sub-circular salient that we call "Ecuadorian Orocline" (Figure 10). CCW rotations between 15°S and 5°S (Peru) were interpreted as marking the northern limb of the Bolivian oroclinal bend (Butler et al., 1995;Gilder et al., 2003;Roperch & Carlier, 1992;Rousse et al., 2005). Our data from the Cordillera Occidental of Ecuador fit in this regional salient-reentrant sequence, as the orogenic trend changes from N20°W to N20°E at ∼4°S, and Neogene CW rotations are consistently observed north of 4°S. In this framework, the along-chain rotation pattern mimics the reentrant-salient sequence of the Andes, and the CCW rotations documented in Peru (15°-5°S) would characterize the shared limb of a Bolivian orogenic reentrant and an Ecuadorian salient ( Figure 10).
By adding further geodynamic suggestions on the formation of Andean orogenic bends, we highlight here the outstanding coincidence between the reentrant-salient chain sequence and the Amazonian Craton outline (Casquet et al., 2012; Figure 11). The inland advance of the Bolivian orocline reentrant coincides with a local lack of stiff cratonic areas east of the accretionary front, whereas the Ecuador salient perfectly encircles the Amazonian Craton boundaries. Thus, the Amazonian Craton may have represented a strong crustal indenter guiding the deformation distribution along the northern Andean belt. We infer that the stiff Archean-Paleoproterozoic core of northeastern South America has forced the Andean orogen to bend around it, whereas the Bolivian Orocline may arise from greater penetration within the continent allowed by a lack of Archean-Paleoproterozoic cratons east of the accretionary front ( Figure 11).

Conclusions
We provide the first paleomagnetic dataset from the Cordillera Occidental, Interandean Valley, and the Cordillera Real of southern Ecuador. The data show a lack of significant strike-slip deformation along the Pallatanga and Peltectec faults after the Early Miocene and Middle Eocene, respectively, strengthening the suggestion that collision of Caribbean-related oceanic terrains with the Ecuadorian margin ended in the Late Cretaceous, at ca. 70 Ma (Luzieux et al., 2006). Eocene-Miocene paleo-inclinations from volcanic data reveal a southern hemisphere provenance but also paleo-latitudes systematically consistent with those expected for the South American plate, arguing firmly against post-Eocene northward terrane displacement in Ecuador.
A remarkably consistent CW rotation pattern of 20°-30° in the Cajas Plateau and the western Interandean Valley, suggests that the Cordillera Occidental emplaced over the non-rotating Cordillera Real as a rotational nappe after the Late Miocene. By considering the age consistency of paleomagnetic rotation and continental deposition onset along the Interandean Valley, we suggest that thrusting is the main cause of tectonic depression genesis. We interpret the Interandean Valley as a narrow flexural basin, formed ahead of a Cordillera Occidental thrust sheet and subsequently incorporated in the deformation, and we suggest that the main rotational thrust front corresponds to the Azogues fault, overlying a non-rotational thrust fault footwall.
The re-evaluation of data by Roperch et al. (1987) and Luzieux et al. (2006) yields a 28° ± 9° CW post-Cretaceous rotation of the Coastal forearc oceanic blocks with respect to South America that is, indistinguishable from the 24° ± 10° CW Neogene rotation documented by us in the Cordillera Occidental and west Interandean Valley. Thus we argue for a Neogene 20°-30° CW rotation of the whole western Andean chain of Ecuador, implying that the Azogues fault is the front of a mid-crustal rotational detachment underlying the entire Cordillera Occidental and Coastal Forearc of southern Ecuador.
Paleomagnetic evidence from Ecuador reveals that the Northern Andean Block-shown to be currently extruded NE-ward by GPS data-is a very recent feature of the northern Andes that has not accumulated a paleomagnetically-detectable displacement.
Finally, we note that the orogenic reentrant-salient sequence of the Nazca trench-Andean chain (and related Cenozoic paleomagnetic rotations) from northern Chile to Ecuador mimics closely the margin of the Archean-Paleoproterozoic Amazonian Craton and other minor cratons of South America. We infer that the stiff crust of the Amazonian Craton behaved as a foreland indenter, hampered inland deformation propagation, and caused the formation of what we call the "Ecuadorian Orocline", arisen by opposite-sign nappe rotations around the Craton apex.

Acknowledgments
The authors thank the Parque Nacional Cajas to allow the paleomagnetic sampling in the natural reserve (research permit N° 206-2019-DPAA/MA; project Paleomagnetismo, representative Maurizio Mulas). The authors also thank the local inhabitants of San Felipe De Molleturo village for the thrilling experience they kindly offered to us. Detailed comments by four anonymous reviewers, the Associate Editor, and the Editor Laurent Jolivet helped us to strengthen data statistical treatment and better focus on the geological implications of paleomagnetic rotations.